INVESTIGATING OFF-RESONANCE FAT MODULATIONS IN THE TURBOSPI SIGNAL TO IMPROVE R2* MAPPING FOR QUANTITATIVE CELL TRACKING

Abstract

Immune cells can be labelled with superparamagnetic iron oxide (SPIO) nanoparticles and detected in vivo. Tracking the migration of immune cells is valuable for understanding the immunogenic response to both cancer and cancer therapies in longitudinal studies. While many sequences are sensitive to SPIO contrast, they have limited specificity and the analysis is solely semi-quantitative. Quantitative cell tracking is better for analyzing immune recruitment, but reports of validated quantification of SPIO labelled cells are rare. This work uses TurboSPI, a multi-echo single point imaging technique that can provide quantification through R2* mapping at high temporal resolution. Since R2* varies linearly with SPIO concentration, R2* maps can be translated into maps of cell density. TurboSPI was initially tested in vivo to assess cytotoxic T lymphocyte (CTL) tracking in response to immunotherapeutics. Analysis revealed that current mono-exponential R2* fitting techniques performed poorly in the presence of fat. Off-resonance signal from fat creates modulations in the signal time course that are detrimental to fitting an accurate R2* decay. In silico methods were used to better understand and account for these fat contributions. We performed Monte Carlo simulations to investigate how the signal time course changes with varying fractions of fat signal. We fit the simulated data using a hybrid Dixon-R2* signal decay model for simultaneous estimation of R2* and fat fraction (ff ). The proposed hybrid fitting technique gives accurate and stable estimates of both ff and R2* across a variety of simulated conditions. Finally, we translated the hybrid technique to real data using in vitro samples of SPIO labelled cells and oil. The data were fit using both the simple mono-exponential decay model and the proposed hybrid technique. The proposed technique gives more stable R2* estimates for SPIO labelled cells when fat is present at fractions greater than 15%. This work represents the first instance of simultaneous R2* and fat estimation using TurboSPI. Improved R2* estimates will lead to more accurate quantitative cell tracking in future in vivo studies.